| { | |
| "best_metric": 11.93097972869873, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-100", | |
| "epoch": 0.004610632117663331, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 4.610632117663332e-05, | |
| "grad_norm": 0.007248171139508486, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 11.9313, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 4.610632117663332e-05, | |
| "eval_loss": 11.930997848510742, | |
| "eval_runtime": 189.7787, | |
| "eval_samples_per_second": 48.119, | |
| "eval_steps_per_second": 6.018, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 9.221264235326664e-05, | |
| "grad_norm": 0.011779400520026684, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 11.9347, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00013831896352989995, | |
| "grad_norm": 0.01022297516465187, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 11.9256, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00018442528470653328, | |
| "grad_norm": 0.009763536043465137, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 11.9332, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00023053160588316657, | |
| "grad_norm": 0.01042894832789898, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 11.9328, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0002766379270597999, | |
| "grad_norm": 0.019424298778176308, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 11.9306, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0003227442482364332, | |
| "grad_norm": 0.013701501302421093, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 11.9262, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00036885056941306655, | |
| "grad_norm": 0.013128082267940044, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 11.9268, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0004149568905896998, | |
| "grad_norm": 0.01037242729216814, | |
| "learning_rate": 1.5e-06, | |
| "loss": 11.9255, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00046106321176633315, | |
| "grad_norm": 0.013505612500011921, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 11.9283, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0005071695329429665, | |
| "grad_norm": 0.011518862098455429, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 11.9287, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0005532758541195998, | |
| "grad_norm": 0.012861590832471848, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 11.9328, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0005993821752962331, | |
| "grad_norm": 0.012600081041455269, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 11.9359, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0006454884964728665, | |
| "grad_norm": 0.014691936783492565, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 11.9251, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0006915948176494998, | |
| "grad_norm": 0.01612277142703533, | |
| "learning_rate": 2.5e-06, | |
| "loss": 11.9303, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0007377011388261331, | |
| "grad_norm": 0.019262641668319702, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 11.9292, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0007838074600027664, | |
| "grad_norm": 0.01421594899147749, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 11.9336, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0008299137811793996, | |
| "grad_norm": 0.021003536880016327, | |
| "learning_rate": 3e-06, | |
| "loss": 11.9341, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.000876020102356033, | |
| "grad_norm": 0.01781291700899601, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 11.9307, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0009221264235326663, | |
| "grad_norm": 0.015007808804512024, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 11.9386, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0009682327447092996, | |
| "grad_norm": 0.01923958584666252, | |
| "learning_rate": 3.5e-06, | |
| "loss": 11.9358, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.001014339065885933, | |
| "grad_norm": 0.01430001575499773, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 11.9321, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0010604453870625664, | |
| "grad_norm": 0.023019419983029366, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 11.9269, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0011065517082391996, | |
| "grad_norm": 0.014341956935822964, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 11.9333, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0011526580294158328, | |
| "grad_norm": 0.017692534253001213, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 11.9276, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0011987643505924662, | |
| "grad_norm": 0.018241120502352715, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 11.9286, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0012448706717690995, | |
| "grad_norm": 0.01712462306022644, | |
| "learning_rate": 4.5e-06, | |
| "loss": 11.9349, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.001290976992945733, | |
| "grad_norm": 0.024579280987381935, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 11.9306, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0013370833141223661, | |
| "grad_norm": 0.014921668916940689, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 11.9297, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0013831896352989996, | |
| "grad_norm": 0.019532358273863792, | |
| "learning_rate": 5e-06, | |
| "loss": 11.9351, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0014292959564756328, | |
| "grad_norm": 0.023204781115055084, | |
| "learning_rate": 4.997482666353287e-06, | |
| "loss": 11.9232, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0014754022776522662, | |
| "grad_norm": 0.02387520670890808, | |
| "learning_rate": 4.989935734988098e-06, | |
| "loss": 11.9375, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0015215085988288994, | |
| "grad_norm": 0.0345429852604866, | |
| "learning_rate": 4.977374404419838e-06, | |
| "loss": 11.9248, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0015676149200055329, | |
| "grad_norm": 0.025636054575443268, | |
| "learning_rate": 4.959823971496575e-06, | |
| "loss": 11.9312, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.001613721241182166, | |
| "grad_norm": 0.02809992991387844, | |
| "learning_rate": 4.937319780454559e-06, | |
| "loss": 11.9266, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0016598275623587993, | |
| "grad_norm": 0.031008386984467506, | |
| "learning_rate": 4.909907151739634e-06, | |
| "loss": 11.9359, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0017059338835354327, | |
| "grad_norm": 0.029894821345806122, | |
| "learning_rate": 4.8776412907378845e-06, | |
| "loss": 11.9259, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.001752040204712066, | |
| "grad_norm": 0.02281125821173191, | |
| "learning_rate": 4.8405871765993435e-06, | |
| "loss": 11.93, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0017981465258886994, | |
| "grad_norm": 0.03086579591035843, | |
| "learning_rate": 4.7988194313786275e-06, | |
| "loss": 11.926, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0018442528470653326, | |
| "grad_norm": 0.024276772513985634, | |
| "learning_rate": 4.752422169756048e-06, | |
| "loss": 11.9231, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.001890359168241966, | |
| "grad_norm": 0.02484738826751709, | |
| "learning_rate": 4.701488829641845e-06, | |
| "loss": 11.9357, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0019364654894185992, | |
| "grad_norm": 0.04550924152135849, | |
| "learning_rate": 4.646121984004666e-06, | |
| "loss": 11.9309, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0019825718105952327, | |
| "grad_norm": 0.03542514890432358, | |
| "learning_rate": 4.586433134303257e-06, | |
| "loss": 11.9254, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.002028678131771866, | |
| "grad_norm": 0.037729743868112564, | |
| "learning_rate": 4.522542485937369e-06, | |
| "loss": 11.9239, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.002074784452948499, | |
| "grad_norm": 0.04585666581988335, | |
| "learning_rate": 4.454578706170075e-06, | |
| "loss": 11.9279, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0021208907741251328, | |
| "grad_norm": 0.05111996829509735, | |
| "learning_rate": 4.382678665009028e-06, | |
| "loss": 11.9454, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.002166997095301766, | |
| "grad_norm": 0.03707089275121689, | |
| "learning_rate": 4.3069871595684795e-06, | |
| "loss": 11.9256, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.002213103416478399, | |
| "grad_norm": 0.06370975077152252, | |
| "learning_rate": 4.227656622467162e-06, | |
| "loss": 11.9364, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0022592097376550324, | |
| "grad_norm": 0.0724567249417305, | |
| "learning_rate": 4.144846814849282e-06, | |
| "loss": 11.9393, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0023053160588316656, | |
| "grad_norm": 0.07536933571100235, | |
| "learning_rate": 4.058724504646834e-06, | |
| "loss": 11.9193, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0023053160588316656, | |
| "eval_loss": 11.930991172790527, | |
| "eval_runtime": 190.8971, | |
| "eval_samples_per_second": 47.837, | |
| "eval_steps_per_second": 5.982, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0023514223800082993, | |
| "grad_norm": 0.010148766450583935, | |
| "learning_rate": 3.969463130731183e-06, | |
| "loss": 11.9311, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0023975287011849325, | |
| "grad_norm": 0.011732972227036953, | |
| "learning_rate": 3.8772424536302565e-06, | |
| "loss": 11.93, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0024436350223615657, | |
| "grad_norm": 0.008542712777853012, | |
| "learning_rate": 3.782248193514766e-06, | |
| "loss": 11.9327, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.002489741343538199, | |
| "grad_norm": 0.009703322313725948, | |
| "learning_rate": 3.684671656182497e-06, | |
| "loss": 11.9283, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0025358476647148326, | |
| "grad_norm": 0.013916654512286186, | |
| "learning_rate": 3.5847093477938955e-06, | |
| "loss": 11.9337, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.002581953985891466, | |
| "grad_norm": 0.012623277492821217, | |
| "learning_rate": 3.4825625791348093e-06, | |
| "loss": 11.9281, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.002628060307068099, | |
| "grad_norm": 0.013886111788451672, | |
| "learning_rate": 3.3784370602033572e-06, | |
| "loss": 11.9325, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0026741666282447322, | |
| "grad_norm": 0.010405468754470348, | |
| "learning_rate": 3.272542485937369e-06, | |
| "loss": 11.9322, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.002720272949421366, | |
| "grad_norm": 0.010533565655350685, | |
| "learning_rate": 3.165092113916688e-06, | |
| "loss": 11.9338, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.002766379270597999, | |
| "grad_norm": 0.015023970045149326, | |
| "learning_rate": 3.056302334890786e-06, | |
| "loss": 11.9285, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0028124855917746323, | |
| "grad_norm": 0.01545638870447874, | |
| "learning_rate": 2.946392236996592e-06, | |
| "loss": 11.93, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0028585919129512655, | |
| "grad_norm": 0.015377589501440525, | |
| "learning_rate": 2.835583164544139e-06, | |
| "loss": 11.9311, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0029046982341278988, | |
| "grad_norm": 0.01387741044163704, | |
| "learning_rate": 2.724098272258584e-06, | |
| "loss": 11.9288, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0029508045553045324, | |
| "grad_norm": 0.014318383298814297, | |
| "learning_rate": 2.6121620758762877e-06, | |
| "loss": 11.9313, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0029969108764811656, | |
| "grad_norm": 0.019190404564142227, | |
| "learning_rate": 2.5e-06, | |
| "loss": 11.9256, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.003043017197657799, | |
| "grad_norm": 0.01856730319559574, | |
| "learning_rate": 2.3878379241237136e-06, | |
| "loss": 11.9317, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.003089123518834432, | |
| "grad_norm": 0.01225450448691845, | |
| "learning_rate": 2.2759017277414165e-06, | |
| "loss": 11.9345, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0031352298400110657, | |
| "grad_norm": 0.01742597296833992, | |
| "learning_rate": 2.1644168354558623e-06, | |
| "loss": 11.9302, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.003181336161187699, | |
| "grad_norm": 0.01626632548868656, | |
| "learning_rate": 2.053607763003409e-06, | |
| "loss": 11.9344, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.003227442482364332, | |
| "grad_norm": 0.01698143593966961, | |
| "learning_rate": 1.9436976651092143e-06, | |
| "loss": 11.9246, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0032735488035409654, | |
| "grad_norm": 0.013797706924378872, | |
| "learning_rate": 1.8349078860833125e-06, | |
| "loss": 11.9347, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0033196551247175986, | |
| "grad_norm": 0.015473989769816399, | |
| "learning_rate": 1.7274575140626318e-06, | |
| "loss": 11.9309, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0033657614458942322, | |
| "grad_norm": 0.017822301015257835, | |
| "learning_rate": 1.6215629397966432e-06, | |
| "loss": 11.9317, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0034118677670708654, | |
| "grad_norm": 0.017122985795140266, | |
| "learning_rate": 1.5174374208651913e-06, | |
| "loss": 11.9321, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0034579740882474987, | |
| "grad_norm": 0.02036653459072113, | |
| "learning_rate": 1.415290652206105e-06, | |
| "loss": 11.9375, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.003504080409424132, | |
| "grad_norm": 0.017380934208631516, | |
| "learning_rate": 1.3153283438175036e-06, | |
| "loss": 11.939, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0035501867306007655, | |
| "grad_norm": 0.0208874624222517, | |
| "learning_rate": 1.217751806485235e-06, | |
| "loss": 11.926, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0035962930517773987, | |
| "grad_norm": 0.02793768234550953, | |
| "learning_rate": 1.122757546369744e-06, | |
| "loss": 11.9347, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.003642399372954032, | |
| "grad_norm": 0.017211785539984703, | |
| "learning_rate": 1.0305368692688175e-06, | |
| "loss": 11.9357, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.003688505694130665, | |
| "grad_norm": 0.022742517292499542, | |
| "learning_rate": 9.412754953531664e-07, | |
| "loss": 11.9328, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.003734612015307299, | |
| "grad_norm": 0.02240605652332306, | |
| "learning_rate": 8.551531851507186e-07, | |
| "loss": 11.9287, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.003780718336483932, | |
| "grad_norm": 0.01831667870283127, | |
| "learning_rate": 7.723433775328385e-07, | |
| "loss": 11.936, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0038268246576605653, | |
| "grad_norm": 0.021101830527186394, | |
| "learning_rate": 6.930128404315214e-07, | |
| "loss": 11.9253, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.0038729309788371985, | |
| "grad_norm": 0.02482517622411251, | |
| "learning_rate": 6.17321334990973e-07, | |
| "loss": 11.9261, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.003919037300013832, | |
| "grad_norm": 0.02878445014357567, | |
| "learning_rate": 5.454212938299256e-07, | |
| "loss": 11.9309, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.003965143621190465, | |
| "grad_norm": 0.034189388155937195, | |
| "learning_rate": 4.774575140626317e-07, | |
| "loss": 11.9306, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.004011249942367098, | |
| "grad_norm": 0.02674616314470768, | |
| "learning_rate": 4.1356686569674344e-07, | |
| "loss": 11.9353, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.004057356263543732, | |
| "grad_norm": 0.02872542478144169, | |
| "learning_rate": 3.538780159953348e-07, | |
| "loss": 11.9199, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0041034625847203654, | |
| "grad_norm": 0.03754565492272377, | |
| "learning_rate": 2.98511170358155e-07, | |
| "loss": 11.9285, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.004149568905896998, | |
| "grad_norm": 0.043331414461135864, | |
| "learning_rate": 2.4757783024395244e-07, | |
| "loss": 11.9281, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.004195675227073632, | |
| "grad_norm": 0.033205412328243256, | |
| "learning_rate": 2.0118056862137358e-07, | |
| "loss": 11.93, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0042417815482502655, | |
| "grad_norm": 0.03685666620731354, | |
| "learning_rate": 1.59412823400657e-07, | |
| "loss": 11.9276, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.004287887869426898, | |
| "grad_norm": 0.04748856648802757, | |
| "learning_rate": 1.223587092621162e-07, | |
| "loss": 11.935, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.004333994190603532, | |
| "grad_norm": 0.03603879362344742, | |
| "learning_rate": 9.00928482603669e-08, | |
| "loss": 11.9439, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.004380100511780165, | |
| "grad_norm": 0.042884595692157745, | |
| "learning_rate": 6.268021954544095e-08, | |
| "loss": 11.9186, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.004426206832956798, | |
| "grad_norm": 0.05015327408909798, | |
| "learning_rate": 4.017602850342584e-08, | |
| "loss": 11.933, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.004472313154133432, | |
| "grad_norm": 0.04760831966996193, | |
| "learning_rate": 2.262559558016325e-08, | |
| "loss": 11.9286, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.004518419475310065, | |
| "grad_norm": 0.06750523298978806, | |
| "learning_rate": 1.006426501190233e-08, | |
| "loss": 11.9305, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0045645257964866985, | |
| "grad_norm": 0.07084526866674423, | |
| "learning_rate": 2.5173336467135266e-09, | |
| "loss": 11.912, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.004610632117663331, | |
| "grad_norm": 0.10393932461738586, | |
| "learning_rate": 0.0, | |
| "loss": 11.9212, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.004610632117663331, | |
| "eval_loss": 11.93097972869873, | |
| "eval_runtime": 190.9735, | |
| "eval_samples_per_second": 47.818, | |
| "eval_steps_per_second": 5.98, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 100, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 23868604416.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |